Instructions to use Comfy-Org/Mage-Flow with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusion Single File
How to use Comfy-Org/Mage-Flow with Diffusion Single File:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
low quality image with turbo model
I am generating images using the ComfyUI workflow template(unpack the subgraph) and the Turbo model.
When generating images without changing the default settings, low-quality images are produced (they are not sharp and the lines look messy). Is there a way to fix this?
Both the int8 model and the bf16 model experience the same problem.
shouldn't Text Encode (Prompt) be used for the non-edit model then?
consider using more steps, and sgm uniform instead of simple
shouldn't Text Encode (Prompt) be used for the non-edit model then?
consider using more steps, and sgm uniform instead of simple
It's not work. But thanks for your help.
Maybe I would keep using other models.
First impression: it's really fast. But nothing special in terms of results.
The issue may just be the models capabilities. Speed just comes at a price. but in other models like boogu i get much better results with base(int8) model+distill lora than the og turbo model (both are better than mage flow, but not perfect) . Not sure this model here is worth the effort or has that available yet.
